Perhaps incorporate Mozilla test suite for data: URIs:
    http://www.mozilla.org/quality/networking/testing/datatests.html
also validate the mediatype part and normalize it by omitting things which
will be the same by default.

Check for compliance with latest RFC:
    http://tools.ietf.org/html/rfc3986
(uri_encode might use the wrong default for 'patn')

Try to integrate support for IRIs:
    http://tools.ietf.org/html/rfc3987

Other schemes:
complete IANA list: http://www.iana.org/assignments/uri-schemes.html
opaquelocktoken - http://www.rfc-editor.org/rfc/rfc4918.txt (Appendix C)
svn
mailto - http://tools.ietf.org/html/rfc2368
info - http://tools.ietf.org/html/rfc4452
prospero - http://tools.ietf.org/html/rfc4157
wais - http://tools.ietf.org/html/rfc4156
tel - http://tools.ietf.org/html/rfc3966
sip and sips - http://tools.ietf.org/html/rfc3261
gopher - http://tools.ietf.org/html/rfc4266
tag - http://tools.ietf.org/html/rfc4151

new IMAP URI RFC:
    http://www.rfc-editor.org/rfc/rfc5092.txt

Other NIDs for URNs, some of which are shown here:
    http://en.wikipedia.org/wiki/Uniform_Resource_Name

Other NIDs which have RFCs:
    ietf - http://tools.ietf.org/html/rfc2648
    publicid - http://tools.ietf.org/html/rfc3151
    uuid - http://tools.ietf.org/html/rfc4122
    sici - http://tools.ietf.org/html/rfc2288 (not really standardised there, but is there a proper RFC?)

check these CPAN bugs against URI module:
   mailto encoding: http://rt.cpan.org/Public/Bug/Display.html?id=24934
   encode: http://rt.cpan.org/Public/Bug/Display.html?id=21640

gopher URI found in the wild, use for testing:
       <URL:gopher://ftp.cavebear.com/00/pub/Ethernet-codes>

Once again provide the stripping of '<URI:...>' and such like.